Skip to content

[https://nvbugs/5505402] [fix] Disable deep_gemm for Qwen3 QKNormRoPEAttention and Linear layers due to accuracy issues#7616

Merged
DomBrown merged 1 commit intoNVIDIA:mainfrom
DomBrown:nvbug/5505402
Sep 10, 2025
Merged

[https://nvbugs/5505402] [fix] Disable deep_gemm for Qwen3 QKNormRoPEAttention and Linear layers due to accuracy issues#7616
DomBrown merged 1 commit intoNVIDIA:mainfrom
DomBrown:nvbug/5505402

Commits

Commits on Sep 8, 2025